Question design can influence the validity and reliability of physical activity (PA) self-report instruments. This study assesses the effect of temporal domain (“days” walked versus “times” walked) on survey questions about walking behavior.
A 2005 California Health Interview Survey (CHIS) sub-sample (n = 6332) reported the number of days or times they walked for leisure or transportation in the past 7 days and the usual time spent per day or per time. Question order was randomized by temporal domain. Minutes walked per week (mean ± SE) and adherence to PA guidelines (≥150 min/wk) were assessed.
Estimates of leisure walking remained stable across temporal domain (days = 71.4 ± 2.5 min; times = 73.4 ± 2.4 min), but transportation walking differed depending on domain (days = 70.4 ± 3.2 min; times = 52.5 ± 2.6 min). Adherence to PA guidelines based on leisure walking was stable across temporal domain (days = 14.9 ± 0.6%; times = 14.9 ± 0.6%), but again differed by domain for transportation walking (days = 10.4 ± 0.6%; times = 7.8 ± 0.5%). A large order effect (number-of-days versus number-of-times asked first) was observed for reports of days walking for transportation (days first = 87.8 ± 2.9 min; times first = 52.3 ± 2.5 min).
Temporal domain influences estimates of self-reported transportation walking behavior. Current efforts to capture PA from both transportation and leisure activities in health research appear to present distinct methodological challenges.
McClain is with the Cancer Prevention Fellowship Program, National Cancer Institute, Bethesda, MD. Grant is with the California Health Interview Survey, University of California, Los Angeles. Willis and Berrigan are with the Applied Research Program, National Cancer Institute, Bethesda, MD.